DeepMind AI Discovers Faster Sorting and Cryptography Algorithms
Google DeepMind's new AI system, AlphaDev, discovers faster methods for sorting data and generating cryptographic hashes, with the new code already integrated into standard C++ libraries.
Google DeepMind introduces AlphaDev, a new artificial intelligence system based on its game-playing technology, to discover faster ways to handle fundamental computing tasks. The AI successfully finds a method to sort items in a list up to 70% faster than the best existing methods and speeds up a key cryptography algorithm by 30%. These improvements address the growing need for software optimization as physical chip limits slow down traditional hardware advancements.
These newly discovered algorithms act as essential building blocks for software, meaning even small speed-ups lead to massive global savings in cost and energy. Because sorting and hashing are among the most widely used subroutines in computing, enhancing their efficiency directly benefits a vast array of applications. Researchers note that this AI-driven approach represents an interesting and innovative way to optimize computing performance.
The practical impact of this breakthrough is already spreading across the tech industry. After rigorous independent vetting, Google DeepMind's sorting algorithms are officially added to the standard C++ programming library, marking the first AI-discovered algorithm to be included and the first update to C++ sorting in over a decade. Additionally, the team makes its improved cryptography algorithms available through Abseil, an open-source collection of C++ code used by millions of developers worldwide.